Intel’s new chip manufacturing technology is its greatest hope for a turnaround — and its biggest risk

Intel’s (INTC) efforts to restore the chip giant to its former glory may hinge on a new manufacturing process the company calls 18A.

Short for 18 angstroms — an angstrom is equal to 0.1 nanometers — 18A is, according to Wall Street analysts and chip industry experts, Intel’s last hope to take back the semiconductor crown from rival TSMC (TSM).

Available later this year, 18A takes advantage of two manufacturing techniques that TSMC isn’t using yet: gate-all-around transistors and backside power. Intel says the technologies will improve its chips’ performance and efficiency.

But Intel isn’t just using 18A to reclaim its place as a leading chipmaker. The company is also banking on using the technology to move in on TSMC’s contract manufacturing business by building 18A-based chips for itself and custom versions for third-party customers.

But that’ll prove difficult. TSMC already builds chips for companies including AMD, Apple, and Nvidia. Intel opened up its manufacturing business to outside clients in 2021 under former CEO Pat Gelsinger. But Wall Street analysts and executives grew exasperated with his strategy and what they saw as unrealistic goals for the business, which lost $13.4 billion in 2024 despite recording a revenue of $17.5 billion.

So far, Amazon and Microsoft have signed on to build their own chips using Intel’s 18A process, with the hope that others will follow suit. But CFO David Zinsner says third-party commitments are “not significant” yet, according to Reuters.

Intel’s 18A is so important because it’s introducing two technologies to the company’s chips at once. First, it’s taking advantage of gate-all-around transistors — next-generation transistors that can more actively control the flow of electricity through a chip. It’s also using a technique called backside power, which changes where and how power is delivered to a chip’s transistors to enable more efficiency and better performance.

Together, the two technologies would help improve AI application performance without running into energy constraints, UC Santa Barbara engineering professor Kaustav Banerjee told Yahoo Finance. That would cut down on problems like overheating — something that reportedly plagued Nvidia’s Blackwell graphics processing units (GPUs) when they were in development.

Taking first-mover advantage in the chip industry typically means big wins for a semiconductor manufacturer. Intel’s founders invented modern semiconductors, and the chipmaker was the first to successfully manufacture a new kind of transistor called Finfet in 2011.

29 April 2025, USA, San Jose: Lip-Bu Tan, Chief Executive Officer of Intel, appears at an event organized by the company. Photo: Andrej Sokolow/dpa (Photo by Andrej Sokolow/picture alliance via Getty Images) — Intel CEO Lip-Bu Tan appears at an event organized by the company in San Jose, Calif. (Andrej Sokolow/picture alliance via Getty Images) · picture alliance via Getty Images

But TSMC flipped the script in 2019 when it became the first to successfully use advanced chipmaking technology called EUV lithography — massive machines worth hundreds of millions of dollars — to make semiconductors, which helped it soar past Intel and make the world’s most advanced AI chips for companies including Apple and Nvidia.

That’s not all. Intel also already had to delay the rollout of 18A from the first half of 2025 to the second half, according to the company’s previous earnings calls.

Trying to perfect both backside power and gate-all-around transistors at once introduces greater manufacturing complexity and greater room for error.

“Both of these technologies [are] enormously complex on [their] own,” Chris Miller, author of “Chip War,” told Yahoo Finance. “So it’s even harder to do them simultaneously.”

A manufacturing employee at Intel’s foundry in Oregon told Yahoo Finance on the condition of anonymity that the technology wasn’t ready for external customers in December. In a follow-up interview in March, however, they said 18A “has gotten a lot better” and Intel staffers are “optimistic” about it.

Another manufacturing employee at the fab said 18A is “on track.” Still, they worried that Intel’s planned layoffs could depress morale and hinder their work toward getting the process out the door.

TSMC isn’t sitting idly by, though. The company is also launching gate-all-around transistors through its N2 technology, which it plans to release later this year. It’s also working to add backside power to its chips in 2026.

Ensuring 18A works is only part of the equation. Intel also has to prove it can sign up customers to take advantage of the technology for their own chips and can pump them out in the volumes they need.

“Can they do it? Yes, they can do it,” Bank of America analyst Vivek Arya said. “But can they do it with the kind of yield and scale like TSMC? That, I think, remains to be seen.”

Gelsinger staked part of Intel’s revival on his plans to turn the company into a third-party foundry. Tan appears to be set to stick to that plan. And while Intel expects its foundry business, which has been hemorrhaging money, to break even by 2027, Wall Street may not be willing to wait that long.

Various analysts have called for Intel to ditch its third-party foundry, and even get out of the chip manufacturing business entirely, and stick to designing semiconductors like rivals AMD and Nvidia.

But since Intel is the US’s sole large-scale advanced chip manufacturer, the government is keen on keeping its manufacturing arm intact. Intel has secured $7.8 billion in US CHIPS Act funding, and ditching its foundry would put that funding at risk.

“The US doesn’t want to be exclusively dependent on foreign firms for advanced production and R&D. Right now, Intel is the only firm with advanced R&D in the United States,” author Chris Miller explained.

While TSMC is expanding its US footprint, investing $165 billion in new plants and research facilities in the coming year, only one-third of its most advanced chipmaking will take place in the US, the company said in an April call with investors.

Still, Bank of America’s Vivek Arya said TSMC’s US expansion “blunts Intel’s advantage to some extent.”

For Intel, it all comes down to proving 18A can succeed. We’ll find out later this year.